Delft University of Technology
Seismic interferometry applied to local fracture seismicity recorded at Planchón-Peteroa
Volcanic Complex, Argentina-Chile
Casas, J. A.; Draganov, D.; Badi, G. A.; Manassero, M. C.; Olivera Craig, V. H.; Franco Marín, L.; Gómez, M.; Ruigrok, E. DOI 10.1016/j.jsames.2019.03.012 Publication date 2019 Document Version
Accepted author manuscript Published in
Journal of South American Earth Sciences
Citation (APA)
Casas, J. A., Draganov, D., Badi, G. A., Manassero, M. C., Olivera Craig, V. H., Franco Marín, L., Gómez, M., & Ruigrok, E. (2019). Seismic interferometry applied to local fracture seismicity recorded at Planchón-Peteroa Volcanic Complex, Argentina-Chile. Journal of South American Earth Sciences, 92, 134-144. https://doi.org/10.1016/j.jsames.2019.03.012
Important note
To cite this publication, please use the final published version (if applicable). Please check the document version above.
Copyright
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons. Takedown policy
Please contact us and provide details if you believe this document breaches copyrights. We will remove access to the work immediately and investigate your claim.
Seismic interferometry applied to regional and teleseismic
events recorded at Planch´
on-Peteroa Volcanic Complex,
Argentina-Chile
Casas, Jos´e Augustoa, Draganov, Deyanb, Badi, Gabriela Alejandrac, Franco, Luisd
a
Facultad de Ciencias Astron´omicas y Geof´ısicas, Universidad Nacional de La Plata, CONICET, Argentina
bDepartment of Geoscience and Engineering, Delft University of Technology, The
Netherlands
cFacultad de Ciencias Astron´omicas y Geof´ısicas, Universidad Nacional de La Plata,
Argentina
dObservatorio Volcanol´ogico de los Andes del Sur (OVDAS-SERNAGEOMIN), Chile
Abstract
The Planch´on-Peteroa Volcanic Complex (PPVC) is located in the Cen-tral Andes, Argentina-Chile. Even though this active volcanic system is considered one of the most dangerous volcano in the region, with more than twenty modest (VEI < 4) Holocene eruptions, knowledge of its subsurface structures, internal processes, dynamics, and their relation, is still limited.
Seismic interferometry (SI) is a high-resolution technique based on anal-yses of the interference of the propagated seismic energy at one or many stations. SI can be used to characterize the subsurface properties of a target area. In particular, previous SI studies performed in the area of the PPVC describe specific ranges of depth; therefore, more information is required for a thorough description of the subsurface features in the area and for a better understanding of the PPVC dynamics.
We apply SI based on autocorrelations of selected regional and tele-seismic events to image the subsurface structures below stations located in
© 2018 Manuscript version made available under CC-BY-NC-ND 4.0 license https://creativecommons.org/licenses/by-nc-nd/4.0/
Argentina and Chile during 2012. The selection of the events is performed according to their location, magnitude, angle of incidence of P-wave seismic energy, and signal to noise ratio in the records. For each station, we extract time windows and we process them using three ranges of frequency, which are sensitive to different ranges of depths.
This work describes depths and zones previously not analyzed in the area. The results not only complement the available geological, geochemical, and geophysical information, but present new information for depths between 5 and ∼400 km depth, increasing the general knowledge of the subsurface features in the PPVC. Finally, we also propose a model for the first 45 km of the subsurface (i.e., down to the Moho), which indicates the crustal structure and the likely distribution of magma bodies in depth.
Keywords:
Planch´on-Peteroa Volcanic Complex, Seismic Interferometry, Regional and teleseismic events, Magma storage in depth
1. Introduction 1
The Planch´on-Peteroa Volcanic Complex -PPVC- (35.223◦S, 70.568◦W; 2
see location inFigure 1) is located in the Andes at the international border 3
between Argentina and Chile. The PPVC is composed of three main volcanic 4
edifices, i.e., the Azufre, the Planch´on, and the Peteroa, out of which the 5
latter is the current active volcano. The PPVC presents overlapped calderas 6
originated from the destruction of several volcanic structures during past 7
explosive events (Tormey,1989). Through analyses of its historical activity 8
and products, this volcanic system is ranked as the most hazardous volcano 9
in Argentina (Elissondo and Far´ıas,2016) and the eighth most risky volcano 10
in Chile (Technical sheet, Observatorio Volcanol´ogico de los Andes del Sur, 11
OVDAS-SERNAGEOMIN, Chile). 12
The knowledge of the PPVC has been developed by the contribution 13
from several disciplines, i.e., geology (Tormey, 1989; Haller et al., 1994; 14
Naranjo et al.,1999;Tapia Silva,2010;Haller and Risso,2011), geochemistry 15
(Benavente, 2010; Tassi et al., 2016; Benavente et al., 2016), meteorology 16
(Guzm´an et al.,2013), ash analysis (Ramires et al.,2013), seismology (Casas 17
et al.,2014;Manassero et al.,2014;Olivera Craig,2017;Casas et al.,2018; 18
Casas et al.), gravimetry (Tassara et al., 2006), and risk analysis (Haller 19
and Coscarella). These studies contribute to the knowledge of the eruptive 20
history and the current subsurface conditions of this volcanic system. Nev-21
ertheless, the dynamics the PPVC and their relation with the subsurface 22
structures are still poorly understood, increasing the local risk (Elissondo 23
and Far´ıas,2016). 24
A description of the subsurface structures (i.e., depth, associated dimen-25
sions, density contrasts, etc.) is essential for developing accurate knowl-26
edge of the dynamics of any volcanic system. In particular, knowledge of 27
subsurface discontinuities provides constraints for tomographic studies, for 28
magma-ascent modeling, among others, contributing to a better inference 29
of the subsurface conditions, and, therefore, leading to more reliable analy-30
ses of likely future volcanic scenarios. Based on structural-geology analyses, 31
Tapia Silva (2010) describes the subsurface geological units located in the 32
very first 10 km of the subsurface in the area of the PPVC, and present 33
their distribution in depth. Even though no local studies have been applied 34
for describing the crustal structure in the PPVC, Far´ıas et al. (2010) and 35
Giambiagi et al. (2012) provide a crustal structure as a function of depth 36
and the distance from the trench in the Central Andes. For the depth of 37
the subducting slab below the PPVC, they estimate four zones delimited in 38
depth at ∼12 (the intracrustal discontinuity), ∼27, and 45 km depth -the 39
crust-mantle discontinuity (the Moho). The Moho is estimated at ∼45 km 40
depth (Tassara et al.,2006); the intra-lithosphere discontinuity (top of litho-41
spheric low-velocity zone), at ∼70 km depth (Karato,2012); and the top of 42
the subducting slab, at ∼120 km depth (Tassara et al.,2006). Nevertheless, 43
more scientific evidence is required to increase the information about the 44
known subsurface structures, leading to a more accurate characterization of 45
their properties, as well as to describe the subsurface features previously not 46
analyzed. These goals motivate local studies, as the one presented in this 47
article. 48
Claerbout (1968) has constituted a frame over which the theory of seis-49
mic interferometry developed. This passive seismic method -from here on, 50
Seismic Interferometry by Autocorrelations (SIbyA)- suggests that the au-51
tocorrelation of a plane-wave transmission response propagating in a hori-52
zontally layered medium, recorded at the surface, allows the retrieval of the 53
reflection response of a virtual source co-located to the recording station. 54
SIbyA has shown to be a robust method; it has been applied to different 55
type of seismic data, in several areas and at different scales. For example, 56
SIbyA was applied to global- and teleseismic phases to imaging the crustal 57
subsurface at regional scales (Ruigrok and Wapenaar,2012;Nishitsuji et al., 58
2016), to P-wave of microseismic events to imaging the shallow volcanic sub-59
surface (Kim et al.,2017), and to ambient-noise seismic data at several scales 60
(Draganov et al.,2007; Gorbatov et al.,2013;Boullenger et al.,2014;Oren 61
and Nowack,2017). The robustness of SIbyA has motivated its application 62
to local (Casas et al.), regional, and teleseismic seismic data recorded in the 63
area of the PPVC. 64
Nishitsuji et al. (2016) apply SIbyA to global seismic phases recorded 65
in the eastern flank of the Peteroa volcano during 2012. They confirm the 66
location of the Moho at ∼45 km depth, and propose a deformation feature 67
in the subducting slab in the form of detachment, shearing, necking, or any 68
combination of them. 69
Casas et al. apply SIbyA to local seismic events to image the subsurface 70
below the stations located in the Argentine and Chilean sides of the PPVC 71
during 2012. They confirm the geological structure described for the first 72
4 km of the subsurface (Tapia Silva, 2010), provide information about re-73
gions of higher heterogeneity caused by faulting and complex geochemical 74
processes, and support the presence of a magma body emplaced at ∼4 km 75
depth (previously suggested byBenavente (2010)). 76
We apply SIbyA to regional and teleseismic events selected according to 77
their location, magnitude, angles of incidence of the P-wave seismic energy 78
at each station, and the signal to noise ratio in the records. The results 79
for three different frequency ranges allow the description of the subsurface 80
structures between ∼5 and 400 km depth, and the inference of the crustal 81
structure and the location of magma bodies down to the Moho. 82
2. Data 83
The present application uses seismic data recorded by stations deployed 84
in Argentina and Chile during 2012 (see station distribution in Figure 1). 85
The temporal deployment of seismic instruments in an area of interest is 86
a widely used tool for reaching several goals, e.g., perform first analyses of 87
the propagated wavefield and the subsurface conditions, increase the num-88
ber of the recording stations, extend the analyzed area, and improve the 89
accuracy of the results. The MalARRgue project (Ruigrok et al.,2012) was 90
designed by institutions from The Netherlands (Delft University of Technol-91
ogy -TUDelft), Argentina (Comisi´on Nacional de Energ´ıa At´omica CNEA), 92
and The United States (Boise State University -BSU). Its goal is imaging 93
and monitoring the subsurface of the Malarge region (Mendoza, Argentina), 94
an area of high scientific interest due to peculiar volcanic and tectonic pro-95
cesses (Stern, 2004). The MarlARRgue project consisted in a temporal 96
deployment (from January 2012 to January 2013) of 38 stations, out of 97
which six were deployed along the eastern flank of the PPVC (from here 98
on, the PV array). The PV array was equipped with short-period (2 Hz) 99
three-component (Sercel L-22) sensors. 100
Another source of data is provided by three broad-band stations of the 101
Observatorio Volcanol´ogico de los Andes del Sur (OVDAS-SERNAGEOMIN, 102
Chile), which are located ∼6 km northwards. These stations (from here on, 103
OVDAS array) were active during 2012, through the same period as the PV 104
array. 105
3. Application and results 106
SIbyA is described by the reciprocity theorem of correlation type ( Wape-107
naar,2003,2004). Based on this theorem for transient sources (Wapenaar 108
and Fokkema,2006), and using autocorrelation in the time domain, we ob-109 tain: 110 X sources T (xA, −t) ∗ si(−t) ∗ T (xA, t) ∗ si(t) ⊗ s(−t) ∗ s(t)i ≈ −R(xA, −t) + δ(t) − R(xA, t) , (1)
which states that the reflection response R(xA, t) can be retrieved at
111
the station A located (at xA) at the surface through the autocorrelation
112
of a recorded transmitted wavefield T (xA, t). The operator ∗ indicates
113
convolution, ⊗ means deconvolution, and δ is the Dirac’s delta. The fac-114
tor s(−t) ∗ s(t)i corresponds to the autocorrelated source time function 115
(ASTF), which allows the deconvolution of each source time function si(t).
116
Even thoughEquation 1requires sources over the whole stationary phase 117
area (i.e., the Fresnel Zone), seismic events present a non-uniform spatial 118
distribution. Therefore, performing a selection of the seismic sources to be 119
used is essential for a proper application of SIbyA. In order the transmission 120
response of the propagated seismic energy to be accurately estimated by 121
the vertical component of the records, we select only seismic events with 122
P-wave seismic energy arriving (sub) vertically to the station at the surface. 123
The retrieved reflection response (from here on, Rv(xA, t)) is related to a
124
seismic source co-located to the station at the surface, radiating P-wave 125
energy vertically downwards. 126
A seismic source in the subsurface release energy that propagates to-127
wards the surface in which it is reflected. This seismic energy is reflected, 128
refracted, and diffracted at the subsurface structures and heterogeneities (or 129
the surface), part of which arrives to the recording station at the surface. 130
Seismograms are then composed of direct waves followed by these reverber-131
ated waves. SIbyA removes the times previous to the direct arrival, and 132
attenuates the incoherent noise, providing seismic evidence of the location 133
of the subsurface structures. Figure 2 pictures the application of SIbyA in 134
an idealized horizontally layered 2-D medium, given a plane wavefield orig-135
inated by a seismic source located exactly below the station. The obtained 136
reflection response can be used to know the depth of the reflectors located 137
in the subsurface below the station. 138
In the real Earth, nor the wave fronts are plane at local and regional 139
scales nor the subsurface is horizontally layered in volcanic zones. In highly 140
heterogeneous zones (as, for example, the area of the PPVC;Manassero et al. 141
(2014)), the location of a seismic source exactly below the station is not an 142
imperative condition for an accurate retrieval of the subsurface reflection 143
response Rv(xA, t) (Fan and Snieder, 2009), i.e., the vertical component
144
of the records is still an accurate estimation of the transmission response. 145
Therefore, sources with small P-wave angles of incidence are selected. 146
3.1. Pre-processing 147
This section aims to get the input data and prepare it for the proper 148
application of theEquation 1. Using the reference seismic catalogs (IRIS and 149
USGS), we select events occurred during the recording period (i.e., January 150
2012 until January 2013) and which are characterized by a sufficiently great 151
magnitude so that signal to noise ratio is high in the records of each station. 152
Due to likely variations of the local seismic wavefield in space and time, we 153
judge the signal to noise ratio of each event at each of the stations. 154
For the selection of seismic events, we use the software JWEED (Java 155
version of Windows Extracted from Event Data) developed by IRIS. Based 156
on restrictions in the origin time, the location, and the magnitude, we pre-157
select events (seeFigure 3). According to their epicentral distance, we clas-158
sify them in two groups. One group is composed of events with epicentral 159
distances between 30◦ and 120◦, and magnitudes greater than Mw. 6; each 160
event of this group guarantees a sufficiently small P-wave ray parameter 161
(< 0.08 s/km) so that seismic energy arrives (sub)vertically at the station, 162
i.e., with incident angles <∼ 25◦ (Kennett et al., 1995). The second group 163
is composed of events with epicentral distances lower than 30◦ and magni-164
tudes greater than Mw. 5. These events present a wide range of possible 165
P-wave angles of incidence. Therefore, we perform an examination analysis 166
on this second group in order to select only those events with at least one 167
P-wave phase smaller than the adopted threshold (i.e., 0.08 s/km). The ray 168
parameters estimated by the regional velocity model ak135 (Kennett et al., 169
1995) are appropriate for this analysis, as smaller angles of incidence of the 170
P-wave energy are expected in the real Earth, provided its higher hetero-171
geneity (Fan and Snieder,2009). Once the seismic events are selected, there 172
is no need to keep the distinction between the groups, i.e., they are equally 173
significant. 174
The origin time of the selected events is used to extract the seismic 175
waveforms from the records of the PV and OVDAS stations. A first estimate 176
of the P- and S-wave arrival times for each event is calculated using the 177
regional velocity model ak135. These times are then employed to manually 178
pick accurate P- and S-wave arrival times, as well as to get the frequency 179
range of a sufficiently high signal-to-noise ratio. We request a good (> 180
0.8) signal to noise ratio for the events to be processed, in order to avoid 181
non-interested high amplitudes. 182
Provided the origin time of the selected events, obtained the accurate 183
arrival times, and examined the (sub)vertical incidence of the P-wave en-184
ergy and high signal-to-noise ratio of the records, we extract the vertical-185
component records of the selected events at each of the used stations. 186
3.2. Processing 187
The vertical-component records of seismic events with P-wave energy 188
arriving vertically at a station represent an accurate estimate of the P-wave 189
transmission response of such propagated wavefield (provided the disconti-190
nuities are not excessively inclined;Nishitsuji et al. (2016)). 191
Out of the frequency range of processing previously selected for each 192
event according to its signal to noise ratio in the results, we use those fre-193
quencies greater than 0.3 Hz, a threshold defined by the instrumental char-194
acteristics of the PV-array stations (Nishitsuji et al., 2014). Furthermore, 195
we only use those frequencies shared through the events, i.e., [0.3 2.1] Hz. 196
In order to perform a better interpretation of the results through depth, we 197
segmented this frequency range in two sub-ranges, i.e., [0.3 0.8] Hz and [0.8 198
2.1] Hz. The separating frequency (0.8 Hz) is selected after a trial and error 199
approach, based on the observed coherency in the results for all the stations 200
in advanced stages of the processing. 201
In order to avoid the rise of non-physical arrivals caused by cross-terms 202
in the correlations, we extract the times between the first P-wave arrival 203
and the first S-phase arrival. As an example,Figure 4shows the processing 204
windows for the station PV04, for the complete range of frequencies (i.e., 205
[0.8 2.1] Hz). 206
The higher value of the selected frequency range (i.e., 2.1 Hz) restricts 207
the resolution of the results for particular depths. Therefore, out of the 208
(previously tested) vertically arriving seismic events, we make a third group 209
composed of those with epicentral distances smaller than 20◦. These events 210
are characterized by a sufficiently high signal-to-noise ratio up to 3.2 Hz. 211
As this group aims to provide information about shallower subsurface struc-212
tures, we select a minimum frequency of 1 Hz. Therefore, we apply the 213
same processing workflow to the three selected frequency ranges, i.e., [0.3 214
0.8] Hz, [0.8 2.1] Hz, and [1 3.2] Hz. As the same importance is assigned to 215
the events of each of the three groups, we normalize the processing windows 216
according to their vertical flux of seismic energy. 217
As suggested by Equation 1, we estimate and deconvolve the ASTF 218
from each of the autocorrelated time windows. The ASTF of each event is 219
estimated by the main lobe and the secondary monotonous-decreasing lobes, 220
as shown inFigure 5 for the station AD2 and the frequency range [0.3 0.8] 221
Hz. 222
Figure 6presents the autocorrelation of the time windows for the station 223
PV01 and the frequency range [0.3 0.8] Hz, in which each trace is decon-224
volved by its previously estimated ASTF. This figure shows the dominance 225
of the main lobe in the autocorrelated deconvolved traces. These features 226
close to 0 s are mainly non-physical amplitudes remaining from the decon-227
volution. Therefore, these amplitudes are removed through windowing. 228
SIbyA is based on the autocorrelation of time windows extracted from 229
the records of selected seismic events. Note that the autocorrelation of 230
a extracted time window could arise non-physical arrivals at times equal 231
to the time interval between two P-wave arrivals, reducing the quality of 232
the results. However, these time intervals are a function of the epicentral 233
distance of the events. The seismic events used in this application present 234
a wide range of epicentral distances, so that the non-physical arrivals are 235
located at different times in the autocorrelations, leading to a destructive 236
interference of their energy during stacking (seeFigure 7). 237
The last step in the application of Equation 1is stacking the resulting 238
seismic traces for each station, which enhances the energy from the sta-239
tionary phase area. Figure 8 shows the pre-stack panel (deconvolved and 240
windowed traces) and the stacked traces for stations AD2 and PV04, for the 241
three selected frequency ranges of processing. 242
4. Interpretation and discussion 243
Aiming to compare the seismic results with the known features of the 244
subsurface, we transform the time vector of the results to depth through 245
construction and utilization of a velocity model. This model is composed of 246
velocities provided by the regional model ak135 for depths greater than 60 247
km, and a modified version of the one obtained by Bohm et al. (2002) for 248
shallower depths (see used velocity model inFigure 9). 249
Figure 10 and Figure 11 show the stacked traces for the PV and OV-250
DAS arrays, respectively, for each processing frequency range. These figures 251
also show the interpreted subsurface features for each of the stations. As a 252
complex impedance contrast through depth is expected for the area of the 253
PPVC, we only seek for the dominant amplitudes on the obtained reflection 254
responses, which are potentially related to the main subsurface discontinu-255
ities. The lower frequency range (i.e., [0.3 0.8] Hz) leads to describe the 256
subsurface between ∼40 and 400 km depth. The results for the other two 257
frequency ranges (i.e., [0.8 2.1] Hz, and [1 3.2] Hz) allow to interpret the 258
subsurface features for depths between 5 and ∼45 km. The minimum depth 259
limit is set by the non-physical amplitudes removed from close to 0 s after 260
deconvolution. The maximum depth limit is set by the coherency in the 261
results for all the frequency ranges and all the used stations. 262
The interpretation of the results for the smallest frequency range ([0.3 263
0.8] Hz) is performed through contrast of the seismic results and the expected 264
location of the known subsurface features based on the geodynamic scenario 265
and the available geological information for the area of the PPVC (Ferr´an 266
and Mart´ınez,1962;Tassara et al.,2006;Benavente,2010;Tapia Silva,2010; 267
Karato,2012). 268
The results for the PV array (see Figure 10a) show six dominant ampli-269
tudes (i.e., local maximum on the absolute values of the waveforms), which 270
we classify as potential subsurface discontinuities. The close location of the 271
identified features in the seismic results and the known subsurface features 272
lead to the interpretation of the Mohorovicic discontinuity at ∼45 km depth, 273
the intra-lithospheric discontinuity at 65 km, the top of the subducting slab 274
between 110 and 120 km, the bottom of the subducting slab between 140 275
and ∼160 km, the lithosphere-asthenosphere boundary between 230 and 255 276
km, and the top of the asthenospheric low-velocity zone between ∼330 and 277
∼360 km depth. 278
The OVDAS array (see Figure 11a) is an array located ∼6 km to the 279
north of the PV array, composed of half the stations of the latter, and 280
with greater longitudinal extension. The results for the OVDAS array al-281
low to interpret the Mohorovicic discontinuity at ∼45 km depth, the intra-282
lithospheric discontinuity between 70 and 90 km, the top of the subducting 283
slab between 115 and 130 km, the bottom of the subducting slab between 284
∼165 and ∼185 km, the lithosphere-asthenosphere boundary at ∼250 km, 285
and the top of the asthenospheric low-velocity zone between ∼310 and ∼350 286
km depth. 287
Based on the seismic velocity values for the depths of interpretation and 288
the frequency range of processing, the resolution of the seismic results is 5 289
km (Widess, 1973). This value leads to interpret that the results for the 290
OVDAS array do not differ substantially from the results of the PV array, 291
what is expected provided the small geological variation in ∼6 km along 292
the north-south direction for the used processing wavelengths. The best 293
correlation in depth is observed for the Mohorovicic discontinuity (43-48 km 294
depth), the lithosphere-asthenosphere boundary (∼245 km), and the top 295
of the asthenospheric low-velocity zone (∼340 km). A small difference in 296
depth is observed for the intra-lithospheric discontinuity and the top of the 297
subducting slab; even though greater depths are observed in the results of 298
the OVDAS stations, these differences would not be significant based on 299
the vertical resolution of the results. A greater difference is observed for the 300
bottom of the subducting slab, i.e., ∼15 km greater for the OVDAS stations. 301
Although a dominant positive arrival is expected at the depth of the 302
Moho, a dominant negative amplitude is retrieved in the results for most 303
of the stations. Based on the retrieved waveforms, we interpret the pres-304
ence of a complex area at ∼40-55 km depth, causing a perturbation of the 305
amplitudes retrieved for these depths, in particular for those related to the 306
Moho. 307
Even though dipping structures in the subsurface restrict the reflection 308
energy arrived at the surface, we clearly recognize the depth of the top and 309
bottom of the subducting slab. Therefore, two hypotheses arise. One hy-310
pothesis suggests a stair-like subduction, according to which the top and the 311
bottom of the oceanic slab present horizontal (or gently inclined) regions; 312
the different depths estimated in the results of the PV and the OVDAS ar-313
rays for the bottom of the subducting slab could be caused by a local change 314
of the thickness of the subducting lithosphere. Nevertheless, this hypothesis 315
would not explain the lack of seismicity at the longitude of the stations and 316
depths of analysis (US Geological Survey;Nishitsuji et al.(2016)). A second 317
hypothesis (Nishitsuji et al.,2016) proposes a slab deformation in the form 318
of detachment, shearing, necking, or any combination. Then, a differential 319
deformation between the latitudes of the PV and OVDAS arrays would ex-320
plain the estimated depths for the bottom of the subducting slab. Finally, 321
more information is required to elucidate the proper interpretation. 322
For the two higher ranges of frequencies (i.e., [0.8 2.1] Hz and [1 3.2] Hz) 323
(seeFigure 10b,Figure 10c,Figure 11b, andFigure 11c), the interpretation 324
is also based on the identification of the dominant amplitudes in the results, 325
and the depths for which the arrived reflected energy is particularly smaller, 326
a feature probably caused by the emplacement of a sufficiently great volume 327
of magma as to be manifested in the seismic results. 328
The results for the PV array and the frequency range [0.8 2.1] Hz (see 329
Figure 10b) indicate five clear dominant arrivals in most of the stations, out 330
of which four are between ∼10 and ∼30 km depth and another one at ∼40 km 331
depth. Additionally, we identify an apparent lack of dominant amplitudes 332
for depths between ∼30 and ∼40 km (indicated with an arrow inFigure 10b). 333
The features identified for [0.8 2.1] Hz are supported by the results for the 334
frequency range [1 3.2] Hz (Figure 10c), which improve the depth of the 335
inferred subsurface discontinuities. In addition, these results manifest an 336
apparent low-amplitude region at ∼25 km depth for the western stations of 337
the array. The results for this frequency range also show a dominant arrival 338
at ∼6 km depth. 339
The results for the OVDAS stations agree with the interpretation per-340
formed for the PV array, for the two analyzed frequency ranges. Therefore, 341
we identify local-maximum amplitudes, as well as apparent small-amplitude 342
zones, at roughly the same depths for the two arrays and for the two higher-343
frequency ranges, even though the effect of attenuation increases for the 344
highest frequencies (around 3 Hz in this application) (Sch¨on,2015). Then, 345
these results allow the interpretation of the subsurface structures between 5 346
and ∼45 km depth (the Moho). 347
Based on the average depth of the reflectors interpreted in the seismic re-348
sults, the available scientific information about the subsurface in the PPVC, 349
the proposed structure of the crust for the Central Andes (Far´ıas et al., 350
2010;Giambiagi et al.,2012), and the physics of magma storage in the crust 351
Jackson et al. (2018), we propose a model for the distribution of magma 352
reservoirs in depth in relation to the main subsurface structures in the crust 353
(see Figure 12). 354
Through comparison of the average depth of the interpreted reflectors 355
below the stations and the proposed structure of the crust (Far´ıas et al., 356
2010;Giambiagi et al.,2012), we associate the interpreted reflectors at ∼12, 357
∼18, and ∼32 km depth as the intra-crustal discontinuity (rigid-ductile dis-358
continuity in the upper crust), the discontinuity between the upper and lower 359
crust, and the rigid-ductile discontinuity in the lower crust, respectively (see 360
Figure 12). 361
Jackson et al.(2018) models the formation, storage, and chemical differ-362
entiation of magma in the Earth’s crust. According to the physics of magma 363
storage, the melt fraction is not homogeneously distributed through depth. 364
A great percentage of melt is located in the very upper part of a reservoir, 365
a low percentage is located through most of the reservoir, while a solid area 366
is present in the lower part. The seismic results are most probably evidence 367
of the solid lower section of the reservoir (Jackson et al., 2018). Therefore, 368
we interpret a region in depth as characterized by a magma emplacement 369
in case two conditions are satisfied: 1. the presence of an area of smaller 370
amplitudes in the seismic results, and 2. it is located above any of the inter-371
preted subsurface reflectors. This circumstance is satisfied for two regions, 372
i.e., a shallower zone located above the rigid-ductile discontinuity in the 373
lower crust (i.e., ∼32 km depth); and a deeper one at ∼35 km depth, above 374
a reflector located at ∼40 km. 375
Even though no amplitude information is available for depths lower than 376
5 km depth (which are removed after deconvolution), a subsurface model for 377
the area (Benavente,2010) proposes a magma emplacement at ∼4 km depth. 378
We identify a reflector at ∼6 km depth, which motivates the incorporation 379
of such magma emplacement in our model. 380
Furthermore, two regions (indicated with a question mark in Figure 12) 381
satisfy only one of the imposed conditions, therefore, their interpretation as 382
regions of magma storage is subjected to extra information. These regions 383
are located above the reflectors interpreted at ∼22 depth and the Moho, 384
for which no apparent smaller amplitudes are observed, probably due to 385
its close location to another feature of the subsurface (upper-lower crust 386
discontinuity and the Moho, respectively), or the resolution of the seismic 387
results are not sufficiently great to recognize a region of limited vertical 388
extension of magma. 389
Our results support the information obtained for the subsurface in the 390
area (Yuan et al., 2006; Ward et al., 2013; Gonz´alez-Vidal et al., 2018) 391
which indicate (although with a limited resolution) low-velocity zones for 392
approximately the same range of depths. They are also consistent with 393
the conceptual model proposed for the area (Benavente, 2010) for depths 394
between 5 and 15 km depth, for which great volumes of magma storage are 395
not expected. 396
Finally, more research (e.g., local seismic velocity -or attenuation- tomog-397
raphy studies) is required to accurately identify the location and dimensions 398
of the regions of magma emplacement. 399
5. Conclusions 400
Even though the Planch´on-Peteroa Volcanic Complex (PPVC) is one of 401
the most hazardous volcanic systems in the Central Andes, knowledge of 402
its internal processes, structures, dynamics, and their relation are still not 403
satisfactorily understood. 404
We apply seismic interferometry by autocorrelations to regional and tele-405
seismic data recorded by nine stations deployed in the area of the PPVC (six 406
in Argentina and three in Chile) during 2012. The events are selected accord-407
ing their location, magnitude, angle of incidence of the P-wave energy, the 408
signal to noise ratio on the results, and the related useful frequency range. 409
In order to perform an appropriate description of the subsurface structures 410
below the stations, we use three frequency ranges ([0.3 0.8] Hz, [0.8 2.1] Hz, 411
and [1 3.2] Hz) which are sensitive to different range of frequencies. 412
The smallest frequency range ([0.3 0.8] Hz) is used to infer the tectonic 413
features, i.e., the Moho (at 43-48 km depth), the intra-lithospheric discon-414
tinuity (∼70 km), the top and bottom of the subducting slab (∼120 and 415
∼150-165 km), the lithosphere-asthenosphere boundary (∼250 km), and the 416
top of the asthenospheric low-velocity zone (∼340 km). The results support 417
the hypothesis of deformation in the form of detachment, searing, and/or 418
necking for the longitude of the used stations. Our results also suggest a 419
higher depth (∼15 km) for the bottom of the subducting slab at the north 420
of the PPVC, likely caused by differential deformation along the latitude 421
direction. 422
Based on the results for the two higher-frequency ranges ([0.8 2.1] Hz 423
and [1 3.2] Hz) and previous geological, geochemical, and geophysical infor-424
mation, we propose a model which describes the structure of the crust and 425
the subsurface regions storaging magma bodies down to the Moho. Three 426
regions of sufficiently great volume of magma emplaced at ∼4 km, ∼28 km, 427
and ∼35 km depth, respectively are indicated. 428
The present work provides valuable information about the subsurface 429
conditions of an active volcanic system -the CVPP. We expect the obtained 430
knowledge to be employed in future research aiming to better understand 431
the dynamics of the CVPP. 432
References 433
Benavente, O., 2010. Actividad Hidrotermal asociada a los Complejos 434
Volc´anicos Planch´on-Peteroa y Descabezado Grande-Quizapu-Cerro Azul, 435
36S y 37oS, Zona Volc´anica Sur, Chile. Universidad de Chile . 436
Benavente, O., Tassi, F., Reich, M., Aguilera, F., Capecchiacci, F., 437
Guti´errez, F., Vaselli, O., Rizzo, A., 2016. Chemical and isotopic fea-438
tures of cold and thermal fluids discharged in the Southern Volcanic Zone 439
between 32.5S and 36S: Insights into the physical and chemical processes 440
controlling fluid geochemistry in geothermal systems of Central Chile. 441
Chemical Geology . 442
Bohm, M., L¨uth, S., Echtler, H., Asch, G., Bataille, K., Bruhn, C., Riet-443
brock, A., Wigger, P., 2002. The Southern Andes between 36 and 40S 444
latitude: Seismicity and average seismic velocities. Tectonophysics . 445
Boullenger, B., Verdel, A., Paap, B., Thorbecke, J., Draganov, D., 2014. 446
Studying CO 2 storage with ambient-noise seismic interferometry: A com-447
bined numerical feasibility study and field-data example for Ketzin, Ger-448
many. Geophysics 80, Q1–Q13. 449
Casas, J.A., Badi, G., Manassero, M., Gomez, P., Draganov, D., Ruzzante, 450
J., 2014. Characterization of Seismo-volcanic Activity in Peteroa Volcano, 451
Central Andes Argentina-Chile. Earth Sciences Research Journal 18, 335– 452
336. 453
Casas, J.A., Draganov, D., Badi, G., Manassero, M.C., Olivera Craig, V., 454
Franco, L., G´omez, M., Ruigrok, E., . Seismic interferometry applied to 455
local fracture seismicity recorded at Planch´on-Peteroa Volcanic Complex, 456
Argentina-Chile. Manuscript under revision . 457
Casas, J.A., Mikesell, T.D., Draganov, D., Lepore, S., Badi, G.A., Franco, 458
L., G´omez, M., 2018. Shallow S-Wave Velocity Structure from Ambient 459
Seismic Noise at Planch´on-Peteroa Volcanic Complex, Argentina-Chile. 460
Bulletin of the Seismological Society of America 108, 2183–2198. 461
Claerbout, J.F., 1968. Synthesis of a layered medium from its acoustic 462
transmission response. GEOPHYSICS . 463
Draganov, D.S., Wapenaar, K., Mulder, W., Singer, J., Verdel, A., 2007. Re-464
trieval of reflections from seismic background-noise measurements. Geo-465
physical Research Letters 34. 466
Elissondo, M., Far´ıas, C., 2016. Volcanic Risk assessment in Argentina, in: 467
Cities on Volcanoes IX, Puerto Varas, Chile. 468
Fan, Y., Snieder, R., 2009. Required source distribution for interferometry 469
of waves and diffusive fields. Geophysical Journal International 179, 1232– 470
1244. 471
Far´ıas, M., Comte, D., Charrier, R., Martinod, J., David, C., Tassara, A., 472
Tapia, F., Fock, A., 2010. Crustalscale structural architecture in central 473
Chile based on seismicity and surface geology: Implications for Andean 474
mountain building. Tectonics 29. 475
Ferr´an, O.L.G., Mart´ınez, M.V., 1962. Reconocimiento geol´ogico de la 476
Cordillera de los Andes entre los paralelos 35 y 38 sur, in: Anales de 477
la Facultad de Ciencias F´ısicas y Matem´aticas, pp. ´ag–19. 478
Giambiagi, L., Mescua, J., Bechis, F., Tassara, A., Hoke, G., 2012. Thrust 479
belts of the southern Central Andes: Along-strike variations in shortening, 480
topography, crustal geometry, and denudation. Bulletin of the Geological 481
Society of America . 482
Gonz´alez-Vidal, D., Obermann, A., Tassara, A., Bataille, K., Lupi, M., 483
2018. Crustal model of the Southern Central Andes derived from ambient 484
seismic noise Rayleigh-wave tomography. Tectonophysics . 485
Gorbatov, A., Saygin, E., Kennett, B.L., 2013. Crustal properties from 486
seismic station autocorrelograms. Geophysical Journal International . 487
Guzm´an, C., Hucailuk, C., Tamasi, M., Mart´ınez Bogado, M., Torres, 488
D., 2013. Anomal´ıas Encontradas en los Par´ametros Registrados en la 489
Estaci´on de Medici´on de la Terma del Volc´an Peteroa, in: Actas de ICES 490
IX, pp. 186–194. 491
Haller, M.J., Coscarella, M., . An´alisis probabil´ıstico del riesgo de erupci´on 492
del volc´an Peteroa mediante la aplicaci´on de mezcla de distribuciones 493
exponenciales. Nat. Hazards Earth Syst. Sci 9, 425–431. 494
Haller, M.J., Ostera, H.A., Pesce, A.H., Gardini, M., Folguera, A., 1994. 495
Vulcanoestratigraf´ıa reciente y eruptividad del volc´an Peteroa, in: Con-496
greso Geol´ogico Chileno, pp. 319–323. 497
Haller, M.J., Risso, C., 2011. La erupci´on del volc´an peteroa (3515’s, 7018’o) 498
del 4 de septiembre de 2010. Revista de la Asociacion Geologica Argentina 499
. 500
Jackson, M.D., Blundy, J., Sparks, R.S., 2018. Chemical differentiation, 501
cold storage and remobilization of magma in the Earth’s crust. 502
Karato, S.I., 2012. On the origin of the asthenosphere. Earth and Planetary 503
Science LettersarXiv:1104.0048v2. 504
Kennett, B.L.N., Engdahl, E.R., Buland, R., 1995. Constraints on seismic 505
velocities in the Earth from traveltimes. Geophysical Journal International 506
. 507
Kim, D., Brown, L.D., ´Arnason, K., ´Agustsson, K., Blanck, H., 2017. 508
Magma reflection imaging in Krafla, Iceland, using microearthquake 509
sources. Journal of Geophysical Research: Solid Earth . 510
Manassero, M., Badi, G., Casas, J.A., Gomez, M., Draganov, D., Ruzzante, 511
J., 2014. Seismic attenuation around Peteroa Volcano, Argentina. Earth 512
Sciences Research Journal 18, 341–342. 513
Naranjo, J.A., Haller, M.J., Ostera, H.A., Pesce, A.H., Sruoga, P., 1999. 514
Geologia y peligros del Complejo Volc´anico Planch´on-Peteroa, Andes del 515
Sur (3515’S), Regi´on del Maule, Chile-Provincia de Mendoza, Argentina. 516
Servicio Nacional de Geologia y Mineria. 517
Nishitsuji, Y., Ruigrok, E., Gomez, M., Draganov, D., 2014. Global-phase 518
H/V spectral ratio for delineating the basin in the Malargue Region, Ar-519
gentina. Seismological Research Letters 85, 1004–1011. 520
Nishitsuji, Y., Ruigrok, E., Gomez, M., Wapenaar, K., Draganov, D., 2016. 521
Reflection imaging of aseismic zones of the Nazca slab by global-phase 522
seismic interferometry. Interpretation 4, SJ1–SJ16. 523
Olivera Craig, V., 2017. Relocation of fracture seismicity in Planch´ on-524
Peteroa Volcanic Complex through optimization of the arrival-times iden-525
tification and joint location techniques. Ph.D. thesis. 526
Oren, C., Nowack, R.L., 2017. Seismic body-wave interferometry using noise 527
autocorrelations for crustal structure. Geophysical Journal International 528
. 529
Ramires, A., Elissonde, A., Trombotto Liaudat, D., 2013. Posibles escenarios 530
de riesgo frente a la ca´ıda de cenizas volc´anicas, en el modelo ganadero 531
de la cuenca alta y media del Rio Grande, Malarg¨ue, Mendoza, in: Actas 532
de IX, pp. 304–320. 533
Ruigrok, E., Draganov, D., Gomez, M., Ruzzante, J., Torres, D., Pumarega, 534
I.L., Barbero, N., Ramires, A., Ganan, A.R.C., van Wijk, K., 2012. 535
Malarg¨ue seismic array: Design and deployment of the temporary array. 536
The European Physical Journal Plus 127, 126. 537
Ruigrok, E., Wapenaar, K., 2012. Global-phase seismic interferometry un-538
veils P-wave reflectivity below the Himalayas and Tibet. Geophysical 539
Research Letters 39. 540
Sch¨on, J.H., 2015. Physical properties of rocks: Fundamentals and principles 541
of petrophysics. volume 65. Elsevier. 542
Stern, C.R., 2004. Active Andean volcanism: its geologic and tectonic set-543
ting. Revista geol´ogica de Chile . 544
Tapia Silva, F.F., 2010. An´alisis estructural del sector occidental de la faja 545
plegada y corrida de Malarg¨ue en el curso superior del r´ıo Colorado de 546
Lontu´e (35 18’y 35o23’s), Regi´on del Maule, Chile. Universidad de Chile 547
. 548
Tassara, A., G¨otze, H.J., Schmidt, S., Hackney, R., 2006. Three-dimensional 549
density model of the Nazca plate and the Andean continental margin. 550
Journal of Geophysical Research: Solid Earth . 551
Tassi, F., Aguilera, F., Benavente, O., Paonita, A., Chiodini, G., Caliro, 552
S., Agusto, M., Gutierrez, F., Capaccioni, B., Vaselli, O., Caselli, A., 553
Saltori, O., 2016. Geochemistry of fluid discharges from Peteroa volcano 554
(Argentina-Chile) in 2010-2015: Insights into compositional changes re-555
lated to the fluid source region(s). Chemical Geology . 556
Tormey, D., 1989. Geologic history of the active Azufre-Planchon-Peteroa 557
volcanic center (3515’S, Southern Andes) with implications for the devel-558
opment of compositional gaps. Asoc. Gel. Arg. Rev , 420–430. 559
Wapenaar, K., 2003. Synthesis of an inhomogeneous medium from its acous-560
tic transmission response. GEOPHYSICS . 561
Wapenaar, K., 2004. Retrieving the elastodynamic Green’s function of an 562
arbitrary inhomogeneous medium by cross correlation. Physical Review 563
Letters 93. 564
Wapenaar, K., Fokkema, J., 2006. Green’s function representations for seis-565
mic interferometry. GEOPHYSICS . 566
Ward, K.M., Porter, R.C., Zandt, G., Beck, S.L., Wagner, L.S., Minaya, E., 567
Tavera, H., 2013. Ambient noise tomography across the Central Andes. 568
Geophysical Journal International . 569
Widess, M.B., 1973. How thin is a thin bed? GEOPHYSICS . 570
Yuan, X., Asch, G., Bataille, K., Bock, G., Bohm, M., Echtler, H., Kind, 571
R., Oncken, O., W¨olbern, I., 2006. Deep seismic images of the Southern 572
Andes. Geological Society of America Special Papers . 573
6. Figures 574
Figure 1: Distribution of the seismic stations used in the present application in relation to the main edifices of the Planch´on-Peteroa Volcanic Complex (PPVC).
Figure 2: Seismic interferometry by autocorrelation applied to vertically arriving energy in a horizontally layered medium. tj represents the two-way travel time between the
station at the surface and the reflector j in the subsurface. The autocorrelation allows the retrieval of a seismogram composed of reflected energy released by a virtual source co-located at the position of the station.
Figure 3: Location of seismic events pre-selected for the application of SIbyA in the area of the PPVC. A triangle indicates the location of the PPVC. Stars show the location of events with epicentral distances less than 30◦and magnitudes Mw > 5. Circles indicate events with epicentral distances greater than 30◦and less than 120◦, and magnitudes Mw > 6.
Figure 4: Processing time windows (P-wave codas) for each of the events selected for PV04 station in the complete range of frequencies, i.e., [0.3 2.1] Hz. Each window is normalized according to its vertical energy flux. Vertical axis indicates propagation time. Each window is composed of a pre-event time (20 s) and the times between the first P-and S-wave arrival times.
Figure 5: Autocorrelated source time functions (ASTFs) estimated for the station AD2 for the frequency range [0.3 0.8] Hz. A shaded area shows the ASTFs in the autocorrelation panel (for graphical purposes, we only show the first 15 s).
Figure 6: Autocorrelated time windows for the station PV01 in the frequency range [0.3 0.8] Hz. The vertical axis indicates two-way travel time. Each seismic trace is deconvolved by its previously estimated source time function.
Figure 7: Cartoon illustrating the attenuation of non-physical arrivals originated in the correlation of a time window with several P-wave arrivals. Stacking seismic traces from events with different epicentral distances enhances features located in phase, so that non-physical arrivals due to several P-wave arrivals are attenuated. Without loss of generality, this figure shows the effect of stacking using time windows of events with different epi-central distances, each of them composed of two P-wave phases. Ti is the time window
of the event i, which contains two P-wave arrivals separated in δti. Operator ∗∗ means
(a) AD2 [0.3 0.8] Hz
(b) AD2 [0.8 2.1] Hz
(d) PV04 [0.3 0.8] Hz
(e) PV04 [0.8 2.1] Hz
(f) PV04 [1 3.2] Hz
Figure 8: Pre-stacking panels and stacked seismic trace for the stations AD2 (a, b, c) and PV04 (d, e, f), for the frequency ranges [0.3 0.8] Hz (a, d), [0.8 2.1] Hz (b, e), and [1 3.2]
Figure 9: Velocity model used to perform the time-to-depth transformation of the seismic results.
(a)
(b) (c)
Figure 10: Interpretation of the results at the stations of the PV array for the three frequency ranges: (a) [0.3 0.8] Hz, (b) [0.8 2.1] Hz, y (c) [1 3.2] Hz. Filled rectangle areas show the local maximum amplitudes, i.e., the interpreted subsurface discontinuities below each station. Rectangles with dashed line borders indicate a higher uncertainty at the identification of a discontinuity. Discontinuities interpreted only in (c) are marked with a small circle in the left bottom corner of each rectangle. Figure 10calso shows the interpreted discontinuities at depths close to those interpreted in (b). Lm represent the
(a)
(b) (c)
Figure 11: Interpretation of the results at the stations of the OVDAS array for the three frequency ranges: (a) [0.3 0.8] Hz, (b) [0.8 2.1] Hz, y (c) [1 3.2] Hz. Filled rectangle areas show the local maximum amplitudes, i.e., the interpreted subsurface discontinuities below each station. Rectangles with dashed line borders indicate a higher uncertainty at the identification of a discontinuity. Discontinuities interpreted only in (c) are marked with a small circle in the left bottom corner of each rectangle. Figure 11calso shows the interpreted discontinuities at depths close to those interpreted in (b). Lm represent the
minimum depth level for the Moho (interpreted in (a)). Arrows indicate zones of likely emplacement of magma. The dashed arrow represents an uncertainty of interpretation
Figure 12: Proposed model of magma emplacement in relation to the structure of the crust down to the Moho in the area of the PPVC. Inverted triangles indicate the longitude of the stations. Thick horizontal lines below the stations show the average depth of the reflectors interpreted in the seismic results. Dashed lines are the interpreted discontinuities (based onFar´ıas et al. (2010) andGiambiagi et al.(2012)) between the different regions of the crust. Arrows show the inferred direction of magma movement. Areas with a question mark inside indicate zones of higher ambiguity in the interpretation.
